Affinity purification contaminants identified by cryo-EM and mass spectrometry

This study identifies human propionyl-coenzyme A carboxylase and the PRMT5:MEP50 complex as unexpected contaminants in affinity-purified proteins from HEK293 cells, demonstrating how cryo-EM and mass spectrometry can reveal these impurities and underscoring the critical need for high resin specificity in structural biology workflows.

The Gist

The Big Picture: A High-Stakes Fishing Trip

Imagine you are a scientist trying to catch a very specific, rare, and wobbly fish (your Target Protein) from a massive, chaotic ocean filled with millions of other creatures (your Cell Lysate).

To catch your fish, you use a special fishing rod with a unique lure (an Affinity Tag). You believe this lure is so specific that only your target fish will bite. You cast your line, pull up the fish, and take it to your lab to take a super-high-definition photo (using Cryo-EM) to see exactly what it looks like.

But here is the twist in this story: The ocean is trickier than you thought. Even with your special lure, you accidentally caught some other creatures that look nothing like your target, but they are so perfectly shaped and rigid that they actually took over your photo session, blurring out the picture of the fish you actually wanted.

The Two "Imposter" Catchers

The researchers tried two different fishing strategies (purification methods) and ran into two different types of "imposter" contaminants.

1. The Strep-Tag Trap: The "Biotin Bandit"

• The Strategy: They used a lure called a Strep-tag. This is like a magnetic hook that grabs onto a specific type of bait called StrepTactin.
• The Problem: Inside human cells, there are four natural proteins that are already "sticky" with a substance called biotin. Think of biotin as a super-strong glue.
• The Mix-up: The StrepTactin resin (the fishing hook) is designed to grab the Strep-tag, but it also grabs anything with biotin because the glue is stronger than the tag.
• The Result: The researchers pulled out a massive, rigid enzyme called hPCC (Propionyl-CoA Carboxylase). It wasn't their target, but because it was so stiff and uniform, the computer processing the photos picked it out easily. It looked like a perfect, symmetrical snowflake, whereas their actual target was a floppy, wobbly jellyfish. The computer got confused and focused on the snowflake, ignoring the jellyfish.

2. The FLAG-Tag Trap: The "Chameleon"

• The Strategy: They tried a different lure called a FLAG-tag. This is a tiny peptide that binds to a specific antibody (a "lock and key" mechanism).
• The Problem: They thought this was foolproof. But, they accidentally caught a complex called PRMT5:MEP50.
• The Mix-up: This contaminant doesn't have the FLAG tag at all. However, its surface is electrically charged in a way that mimics the FLAG tag. It's like a chameleon that changes its color to match the lure. The antibody grabbed onto it by mistake, thinking it was the target.
• The Result: Just like the first time, this contaminant was a rigid, well-structured machine. It dominated the data, making it look like the researchers had successfully purified a perfect machine, when in reality, they had purified a "look-alike" intruder.

Why Did This Happen? (The "Rigid vs. Wobbly" Analogy)

You might wonder: "If the target protein was there, why didn't the computer see it?"

Imagine you are trying to take a group photo of a bunch of people.

• The Target Protein: It's a group of people doing yoga, stretching, and moving around constantly. They are all in slightly different poses.
• The Contaminants: They are a group of statues. They are perfectly still and identical.

When the computer tries to stack thousands of photos to create a clear image (a process called averaging), it gets confused by the moving yoga people. But the statues? They are identical in every single photo. The computer thinks, "Wow, these statues are the main subject! Let's focus on them!"

Because the contaminants were so structurally perfect and rigid, they "hijacked" the data analysis, even though there were far fewer of them than the actual target protein.

The Takeaway: Even the Best Tools Have Flaws

The main lesson of this paper is a warning to all scientists: Affinity purification is powerful, but it's not magic.

• Strep-tags can accidentally grab natural biotinylated proteins (like hPCC).
• FLAG-tags can accidentally grab proteins that just look like the tag (like PRMT5).

These contaminants are often invisible in standard tests (like Western Blots) because they don't show up as "wrong" until you try to take the high-resolution photo.

The Solution?

The researchers found a few ways to fix this:

1. Block the Biotin: For the Strep-tag problem, they tried adding a "blocker" (avidin) to cover up the natural biotin so the resin wouldn't grab it. (Though this caused other issues like clumping).
2. Double Purification: The best fix was to use both purification methods in a row. Catch the fish with the Strep-tag, wash it, and then catch it again with the FLAG-tag. This filters out the imposters, though it means you lose a few of your target fish in the process.

In short: When you are looking for a needle in a haystack, sometimes you accidentally pick up a shiny piece of plastic that looks like a needle. You have to be very careful to check what you've actually caught before you start taking pictures!

Technical Summary

1. Problem Statement

Affinity chromatography is a cornerstone of protein purification for structural biology, particularly for cryo-electron microscopy (cryo-EM), due to its ability to isolate specific targets from complex lysates in a single step. However, the authors highlight a critical, often overlooked issue: endogenous cellular proteins can non-specifically bind to popular affinity resins, co-purifying with the target protein.

While these contaminants may be present in low abundance relative to the target, their structural homogeneity allows them to dominate cryo-EM datasets during single-particle analysis (SPA). Unlike flexible or heterogeneous target complexes, rigid contaminants form high-resolution 3D reconstructions from very few particles, potentially misleading researchers into analyzing the wrong structure or wasting significant resources.

2. Methodology

The study utilized a multi-modal approach to identify and characterize contaminants arising from two distinct affinity purification strategies used to isolate a tetrameric TGF-β receptor complex (ALK4/ACVR2A/Activin A) expressed in HEK293F cells.

• Target System: A complex of ALK4 (fused to a twin Strep-tag) and ACVR2A (fused to YPet and a FLAG-tag).
• Purification Strategies:
  1. StrepTactin Resin: Targeted the Strep-tag on ALK4.
  2. Anti-FLAG M2 Resin: Targeted the FLAG-tag on ACVR2A.
• Cryo-EM Data Collection & Processing:
  • Samples were prepared on cryo-EM grids and imaged on Talos Arctica (200 kV) and Titan Krios (300 kV) microscopes.
  • Data processing involved "on-the-fly" preprocessing (Warp) and standard SPA workflows (CryoSPARC).
  • Key Observation: The authors noticed unexpected 3D volumes with high symmetry and high resolution emerging from small particle counts (e.g., ~3,000 particles), which did not match the expected size or symmetry of the target complex.
• Identification Techniques:
  • Visual Inspection: Comparing reconstructed 3D volumes against known structures in the EMDB/PDB.
  • Mass Spectrometry (LC-MS/MS): Proteomic analysis of elution fractions to confirm protein identity.
  • Computational Modeling: Using AlphaFold3 to model potential interactions between the anti-FLAG antibody and the identified contaminant.
• Mitigation Experiments: Testing the addition of avidin to block biotin-binding sites during StrepTactin purification.

3. Key Results

A. Contaminant 1: Human Propionyl-Coenzyme A Carboxylase (hPCC)

• Source: Purification via StrepTactin resin.
• Identification: A 3D volume with D3 symmetry was reconstructed from only 3,097 particles at 6.23 Å resolution. This matched the known structure of hPCC (EMD-33729).
• Mechanism: hPCC is one of four endogenously biotinylated proteins in human cells. StrepTactin binds biotin with extremely high affinity (near-irreversible). Consequently, endogenous hPCC competed with the Strep-tagged target for resin binding.
• Outcome: Despite being less abundant than the target, hPCC formed rigid, homogeneous particles that were preferentially picked during data processing.
• Mitigation: Adding avidin to the sample prior to resin incubation successfully blocked biotinylated contaminants. However, this caused precipitation of the target complex and increased costs, rendering it impractical for this specific workflow.

B. Contaminant 2: PRMT5:MEP50 Complex

• Source: Purification via Anti-FLAG M2 resin.
• Identification: A 3D volume with D2 symmetry was reconstructed from ~9,200 particles at 4.53 Å resolution. Mass spectrometry identified this as the Protein Arginine Methyltransferase 5 (PRMT5) complexed with Methylosome Protein 50 (MEP50).
• Mechanism: PRMT5:MEP50 does not contain a FLAG tag. The authors hypothesize that the highly charged surface of the PRMT5:MEP50 complex mimics the charged FLAG peptide (DYKDDDDK), leading to non-specific electrostatic interactions with the anti-FLAG M2 antibody.
• Validation: AlphaFold3 modeling suggested a potential non-linear surface epitope interaction between the anti-FLAG Fab and PRMT5, though this interface was not reliably predicted (high PAE) and remains speculative. Notably, PRMT5 was not detected in Western blots, likely because the epitope is conformational and lost during SDS-PAGE denaturation.

4. Key Contributions

• Discovery of Specific Contaminants: The paper definitively identifies hPCC and PRMT5:MEP50 as major contaminants in Strep-tag and FLAG-tag purifications from mammalian cells, respectively.
• Cryo-EM as a Diagnostic Tool: The study demonstrates that cryo-EM data processing itself can serve as a powerful quality control step. The presence of "too good to be true" high-resolution maps from low particle counts is a red flag for contamination.
• Mechanistic Insight: It elucidates that biotinylated endogenous proteins are a significant risk for StrepTactin purifications in mammalian systems (often overlooked in product manuals) and suggests a charge-based mimicry mechanism for anti-FLAG resin contamination.
• Workflow Optimization: The authors show that combining orthogonal affinity steps (Strep followed by FLAG) can eliminate contaminants, though it may reduce overall yield due to competition in the first step.

5. Significance

• Impact on Structural Biology: As cryo-EM becomes the method of choice for "difficult" targets (membrane proteins, flexible complexes), sample purity is paramount. This paper warns that even "highly specific" commercial resins can fail in mammalian systems, leading to the misidentification of structures or wasted data collection time.
• Reproducibility and Rigor: The findings urge the community to be vigilant about endogenous contaminants. Researchers should not assume that a single affinity step yields a pure sample, especially when working with mammalian lysates.
• Future Directions: The authors advocate for the development of next-generation affinity tools (e.g., nanobody-based resins, HaloLink) that offer higher specificity and do not cross-react with common cellular components.

In summary, this paper serves as a critical cautionary tale for structural biologists, illustrating how endogenous proteins can hijack affinity purification workflows and dominate cryo-EM datasets, necessitating rigorous validation of sample composition beyond standard Western blotting.